That ruling, Reuters reported Tuesday (June 24), is a victory for the AI company and its industry.
According to the report, U.S. District Judge William Alsup found that Anthropic made “fair use” of books by writers Andrea Bartz, Charles Graeber and Kirk Wallace Johnson in training its Claude large language model (LLM).
However, Alsup also ruled that Anthropic’s copying and storage of more than 7 million pirated books in a “central library” violated the authors’ copyrights and was not fair use, and ordered a trial in December to decide how much Anthropic owes for the infringement.
The Reuters report noted that U.S. copyright law holds that willful copyright infringement can justify statutory damages of up to $150,000 per work.
A spokesperson for Anthropic told Reuters the company was pleased that the court recognized its AI training was “transformative” and “consistent with copyright’s purpose in enabling creativity and fostering scientific progress.”
The writers filed the proposed class action against Anthropic last year, contending the Amazon and Google-backed company used pirated versions of their books without their consent or compensation to teach Claude to reply to human prompts.
The news comes days after the BBC threatened legal action against AI search engine Perplexity, alleging that the company’s “default AI model” was trained using the network’s material. The BBC has demanded that Perplexity end all scraping of its content, delete any copies used for AI development, and propose compensation for the alleged infringement.
A report by the Financial Times noted that this is the first time the BBC has sought legal recourse over content scraping by AI firms, a sign of the mounting concerns that its freely available public sector content is being widely repurposed without authorization.
The broadcaster claims that parts of its content have been reproduced verbatim by Perplexity, with links to BBC articles surfacing in search results, including material that was only recently published online.
BBC executives maintain that such practices harm the BBC’s reputation for impartial journalism and hurt public trust, pointing to internal research that found 17% of Perplexity responses using BBC sources had significant inaccuracies or missing context.
Recent coverage by PYMNTS has spotlighted the rising friction between generative AI companies and publishers over content scraping.
PYMNTS has written about similar legal threats from The New York Times and Dow Jones, as well as the debut of Perplexity’s publisher revenue-sharing program in response to increasing industry backlash.
